The Agent Skills Directory

[SAFE]: The core functionality of the skill focuses on local context optimization and token budgeting. The instructions contain explicit safety guardrails that prevent the agent from compromising code quality or security for the sake of token savings. All cost calculations, metrics tracking, and context pruning operations are performed locally.
[COMMAND_EXECUTION]: Scripts such as relevance_scorer.py, session_checkpoint.py, and session_hook_generator.py execute shell commands via subprocess.run to interact with Git and check system status. These calls are restricted to project metadata retrieval and status checking using safe execution patterns.
[EXTERNAL_DOWNLOADS]: The skill uses standard GitHub URLs for source code hosting and installation, following established patterns for agent skill distribution. No suspicious external dependencies or remote script downloads from untrusted sources are present.

savethetokens